The comparison of the dimensionality results provided by the automated item selection procedure and DETECT analysis

Authors : Ezgi MOR DİRLİK, Seval KARTAL

Pages : 808-830

Doi:10.21449/ijate.1059200

View : 12 | Download : 5

Publication Date : 2022-12-22

Article Type : Research Paper

Abstract :The dimensionality is one of the most investigated concepts in the psychological assessment, and there are many ways to determine the dimensionality of a measured construct. The Automated Item Selection Procedure insert ignore into journalissuearticles values(AISP); and the DETECT are non-parametric methods aiming to determine the factorial structure of a data set. In the current study, dimensionality results provided by the two methods were compared based on the original factorial structure defined by the scale developers. For the comparison of the two methods, the data was obtained by implementing a scale measuring academic dishonesty levels of bachelor students. The scale was conducted on junior students studying at a public and a private university. The dataset was analyzed by using the AISP and DETECT analyses. The “mokken” and “sirt” packages on the R program were utilized for the AISP and DETECT analyses, respectively. The similarities and differences between the findings provided by the methods were analyzed depending on the original factor structure of the scale verified by the scale developers.
Keywords : Mokken scaling DETECT, Automated item selection procedure, Non parametric item response theory, Dimensionality